Abstract
The homeobox gene product Nkx 6.1 is of unknown function but is expressed in the pancreas and the antropyloric mucosa of the stomach. In the adult pancreas, Nkx 6.1 possesses an insulin cell-restricted distribution, whereas its localization in the stomach is unknown. We now show that the vast majority of serotonin-producing enterochromaffin cells of the antropyloric mucosa contain Nkx 6.1-immunoreactive nuclei. In addition, a subpopulation of cells co-storing serotonin and gastrin display Nkx 6.1-positive nuclei. Such cells have been postulated to represent precursors of mature gastrin and serotonin cells. The nuclei of the co-storing cells have previously also been found to be positive for another homeodomain protein, Pdx-1. Pdx-1-deficient animals were therefore investigated and were found to be devoid of Nkx 6.1-positive nuclei. Our data show that Pdx-1 is needed for Nkx 6.1 expression and suggest a role for Nkx 6.1 in the maturation of gastrin- and serotonin-positive precursor cells.
T
It is therefore of interest to determine which factors control the expression of gastrointestinal hormones. We have recently identified two homeobox proteins, isl-1 and Pdx-1, as being associated with endocrine cells of the gastric mucosa (Larsson et al. 1995,1996). Moreover, targeted deletion of the Pdx-1 gene in mice results in animals with nearly total ablation of gastrin cells (Larsson et al. 1996). Interestingly, both isl-1 and Pdx-1 were initially discovered in the pancreatic islets, and Pdx-1 constitutes an important regulator of both pancreatic development (Jonsson et al. 1994) and insulin gene transcription (Ohlsson et al. 1993; Peers et al. 1994; Petersen et al. 1994; Serup et al. 1995,1996). Search for other homeobox gene products that could be involved in islet cell regulation led to the discovery of Nkx 6.1 (Rudnick et al. 1994), which shows a very insulin cell-specific localization in the adult pancreas (Jensen et al. 1996). Moreover, in the early developing pancreas, Nkx 6.1 immunoreactivity is present in most pancreatic parenchymal cells. A similar localization was found for Pdx-1, indicating that both homeodomain proteins may play important roles during early pancreatic differentiation (Øster et al. 1998). During the course of studies of Nkx 6.1, it was found that mRNA encoding this otherwise remarkably insulin cell-specific protein also occurred in the antropyloric mucosa of the stomach (Jensen et al. 1996). We now report that the majority of Nkx 6.1-positive antropyloric nuclei belong to serotonin-producing enterochromaffin cells that may develop from gastrin- serotonin-positive precursor cells (cf. Larsson et al. 1996). In addition, the effect of Pdx-1 deficiency on antropyloric Nkx 6.1-positive nuclei was evaluated.
(A,B) Double immunofluorescence staining for Nkx 6.1 in antropyloric mucosa of adult mouse. Note the scattered Nkx 6.1-immunoreactive nuclei (
Materials and Methods
Animals
Pdx1-deficient mice were generated by deletion of exon 2, which encodes the homeodomain of Pdx-1 (ipf1), using homologous recombination in embryonal stem cells as described by Jonsson et al. (1994). Mice homozygous for the deletion and wild-type (control) mice were sacrificed on the day of birth. The absence of a pancreas was verified in the homozygotes, and specimens from the antropyloric region of the stomach were fixed by immersion for 7 days in 4% paraformaldehyde, dehydrated, and paraffin-embedded, or soaked in 30% sucrose and prepared for cryostat sectioning. Antropyloric stomach with adjoining proximal duodenum and pancreas from rats and mice at different ages, adult, newborn (P1), and P22-P27, were fixed in 4% paraformaldehyde for 20 hr or 7 days and prepared for cryostat sectioning or paraffin embedding.
Immunocytochemistry
Cryostat or paraffin sections of 3-5 μm m were pretreated by three 5-min cycles in 10 mM citrate buffer, pH 6.0, at full power setting (780 W) in a Miehle PP-780 laboratory microwave oven (Miehle & Cie; Gütersloh, Germany) and were submitted to indirect immunofluorescence staining using rabbit antiserum to a recombinant glutathione-S-transferase (GST) fusion protein incorporating the C-terminal region of rat Nkx 6.1 (Ab.174) (Jensen et al. 1996). The recombinant protein was produced in bacteria and purified by affinity chromatography on a glutathione column. The antiserum produced staining in both rat and mouse tissues. In addition, sections were stained with goat anti-serotonin (Incstar; Stillwater, MN), rabbit anti-gastrin (DAKO A/S; Glostrup, Denmark), and monoclonal mouse anti-somatostatin (SOM018, NovoClone; Novo-Nordisk, Bagsvaerd, Denmark) antibodies. For double and triple immunofluorescence, mixtures of primary antibodies raised in different species were used for staining and detection was achieved by species-specific secondary antisera including biotinylated donkey anti-mouse Ig (Jackson Immunoresearch Labs; West Grove, PA) followed by aminomethylcoumarine (AMCA)-conjugated streptavidin (Vector Laboratories; Burlingame, CA), Texas red-conjugated donkey anti-rabbit Ig (Jackson), and fluorescein isothiocyanate (FITC)-conjugated donkey anti-goat Ig (Jackson). To ensure complete species specificity, the secondary antibodies were absorbed overnight with 10 μl/ml normal serum from the species recognized by the other antisera and with 10 μl/ml of normal serum from the species of the specimen.
For some multiple stainings, sections were initially stained for Nkx 6.1 using the peroxidase-anti-peroxidase (PAP) method of Sternberger (1979) and subsequently underwent double or triple immunofluorescence staining for the hormones (serotonin, somatostatin, and gastrin) as described above. Because the Nkx 6.1 stains were nuclear and the hormone stains were cytoplasmic, no interference between immunofluorescence and PAP stains occurred. Sections used for PAP staining were first incubated in methanol-H2O2 to quench endogenous peroxidase activity and were stained with PAP as described (Larsson 1988), using diaminobenzidine-H2O2 medium for development. Sections were examined by both fluorescent epi-illumination (using selective FITC, Texas red, and AMCA filter blocks) and by transillumination using white light. Nkx 6.1-positive nuclei were evaluated by counting the number of positive nuclei in eight Pdx-1-deficient and seven control (wild-type) mice, using eight perfectly transverse sections per animal. Immunocytochemical controls included absorptions of the primary antisera with synthetic gastrin-17 and synthetic somatostatin-14 (Peninsula Laboratories; Merseyside, UK) and with recombinant rat Pdx-1 or Nkx 6.1 cleaved from the GST fusion protein. Serotonin (Sigma; St Louis, MO) was coupled to BSA using formaldehyde as described (Steinbusch et al. 1978) before being used for absorption controls. In addition, conventional staining controls were employed as recommended (Larsson 1988).
Multiplex RT-PCR
cDNA and multiplex reverse transcriptase polymerase chain reaction (RT-PCR) was performed as described using previously published primer sequences (Jensen et al. 1996; Øster et al. 1998). The presence of gastrin, somatostatin, and Nkx 6.1 transcripts and of mRNA encoding the related homeobox protein Gtx (Komuro et al. 1993) was studied in antropyloric mucosa from adult rats and from newborn Pdx-1-deficient and control mice. As internal standards for Nkx 6.1 and Gtx, Sp-1 mRNA was used. Glucose-6-phosphate dehydrogenase mRNA served as standard for gastrin and somatostatin transcripts. Quantitation of product yield was performed by a Molecular Dynamics Series 400 phosphorimager (Jensen et al. 1996).
Results
The Nkx 6.1 antiserum stained scattered nuclei in the epithelium of the antropyloric glands of the adult mouse and rat gastric mucosa (Figure 1A), but revealed no positive nuclei in the duodenum. Absorption of the antiserum with recombinant Nkx 6.1, but not with recombinant Pdx-1, abolished all staining. Similar results were obtained in mouse and rat stomachs. Therefore, in the following, the description refers to both species.
Staining for nuclear Nkx 6.1 immunoreactivity and for the gastric hormones gastrin, somatostatin, and serotonin permitted identification of the Nkx 6.1-positive cells as serotonin-immunoreactive enterochromaffin cells (Figures 1A and 1B). The majority of cells (<90%) with Nkx 6.1-positive nuclei were serotoninimmunoreactive, whereas the remaining were negative for serotonin, gastrin, and somatostatin. The majority (<90%) of antropyloric serotonin cells were also Nkx 6.1-positive. All serotonin cells in the duodenum, as well as serotonin-containing mast cells, were Nkx 6.1-negative. No somatostatin cells displayed Nkx 6.1-positive nuclei, but a small subpopulation (<1%) of gastrin-immunoreactive cells possessed Nkx 6.1-positive nuclei. These cells were invariably also serotoninimmunoreactive. Two different subtypes of gastrin-serotonin double-positive cells could be defined by their nuclear Nkx 6.1 reactivity. One population was characterized by strong gastrin and weaker serotonin immunoreactivity and no nuclear staining for Nkx 6.1 (Figures 1C-1F), and the other was characterized by weaker gastrin immunoreactivity, strong serotonin immunoreactivity and nuclear Nkx 6.1 staining (Figures 1G-1I). Somatostatin cells co-storing serotonin were also detected, albeit infrequently. These cells were Nkx 6.1-negative.
In newborn mice, only very few Nkx 6.1-positive nuclei were seen. Double staining showed that, in newborns, most serotonin cells were Nkx 6.1-negative. Antropyloric mucosa from Pdx-1-deficient and control (wild-type) mice were stained in parallel for Nkx 6.1. No Nkx 6.1-positive nuclei were present in the antropyloric mucosa of any of the eight Pdx-1- deficient mice investigated (eight sections per animal; in all, 99 mm length of mucosa studied), whereas Nkx 6.1-positive nuclei were observed in each of the seven wild-type mice studied (mean ± SD 0.17 ± 0.09 nuclei/mm length of mucosa; eight sections /animal; in all, 89 mm mucosa studied). This difference was significant (p = 0.0006, Mann-Whitney U-test).
Multiplex RT-PCR revealed that no gastrin mRNA could be detected in the Pdx-1-deficient mice, whereas it was present in wild-type mice. Somatostatin and Gtx mRNA levels were the same in both types of animals, but Nkx 6.1 mRNA levels could not be detected. Analyses of adult rat antropyloric specimens revealed that both Nkx 6.1 and Gtx were present at comparable levels.
Discussion
In gastric glands, cell renewal takes place in the regenerative zone (isthmus) by division of a single stem cell. Daughter cells migrate upwards to form surface epithelium and downwards to differentiate into exocrine and endocrine cells. In the antropyloric mucosa, three main types of endocrine cells occur: gastrin, somatostatin, and serotonin cells. We have previously shown that presumptive endocrine stem cells in the isthmus are characterized by co-expression of the gastrin and somatostatin genes (Larsson et al. 1995) (Figure 2). These cells are postulated to undergo asymmetrical cell divisions, resulting in gastrin+ serotonin + and somatostatin + serotonin + daughter cells (Larsson et al. 1996). Part of the gastrin + serotonin + and somatostatin + serotonin + cells display nuclear reactivity for Pdx-1. Mice deficient in Pdx-1 are virtually devoid of gastrin cells, have normal somatostatin cell numbers, and have increased serotonin cell numbers (Larsson et al. 1996). It was therefore postulated that Pdx-1 directs the maturation of gastrin + serotonin + cells into gastrin + cells (Larsson et al. 1996).
Schematic illustration of our present concept of antropyloric endocrine cell differentiation, using data derived from the present study and from Larsson et al. (1996). Gastrin- and somatostatin- positive cells in the isthmus region begin to express Pdx-1 and subsequently undergo asymmetrical cell division, leading to gastrin1serotonin1 and somatostatin1serotonin1 cells. These daughter cells have entered the gland proper and are characterized by Pdx1-positive nuclei (Larsson et al. 1996). The gastrin1 serotonin1 cells may differentiate into either a gastrin or a serotonin cell lineage. In conjunction with lowered gastrin expression (indicated by parentheses), Nkx 6.1 becomes detectable in the nuclei of the double positive cells. With continued development, first gastrin and then Pdx-1 becomes extinguished in the pedigree. Somatostatin1serotonin1 cells are also believed to make a decision for a somatostatinergic or serotoninergic differentiation pathway. Nkx 6.1-positive nuclei are never detected in these cells, and we propose that they constitute precursors of the Nkx 6.1-negative serotonin cells observed in this study.
Our present results document the presence of Nkx 6.1 immunoreactivity in the majority of antropyloric serotonin cells. Becauase Nkx 6.1 shows sequence similarities to another homeodomain protein, Gtx (Komuro et al. 1993), we were concerned about possible crossreactivity to this protein. However, in Pdx-1-deficient mice, we found that Nkx 6.1-immunoreactive nuclei are eliminated, whereas the expression of Gtx is unchanged. Conversely, in adult pancreatic islets, Nkx 6.1 but not Gtx expression occurs and the insulin cell nuclei stain positively with the present antiserum (Øster et al. 1998). Absorption controls and staining controls further attest to the specificity of the Nkx 6.1 staining in both pancreatic islets and stomach. Accordingly, we conclude that our antiserum does not recognize Gtx. Interestingly, triple stainings showed that some cells immunoreactive for both gastrin and serotonin were also Nkx 6.1-immunoreactive. These cells were characterized by weak gastrin and strong serotonin immunoreactivity. Conversely, cells strongly immunopositive for both gastrin and serotonin were Nkx 6.1-negative. This suggests that gastrin+ serotonin + precursor cells display only Nkx 6.1-positive nuclei when the gastrin expression is diminishing. Interestingly, both Nkx 6.1 and Gtx contain consensus motifs characterizing transcriptional repressors (cf. Komuro et al. 1993), which might indicate a role for Nkx 6.1 in reducing gastrin gene expression in cells destined for the serotonin lineage.
The presence of Nkx 6.1-positive nuclei in the gastric endocrine cell pedigree enables us to elaborate on our scheme (Larsson et al. 1996) of gastric endocrine cell development as illustrated in Figure 2.
Because Pdx-1-deficient mice are deficient in gastrin cells but contain increased numbers of serotonin cells (Larsson et al. 1996), the behavior of Nkx 6.1 in such animals was of interest. Regrettably, Pdx-1-deficient mice die shortly after birth (Jonsson et al. 1994), so the study had to be restricted to newborn animals. At this stage, only extremely few Nkx 6.1-positive nuclei are present. By systematic counting of many sections from wild-type and Pdx-1-deficient mice, we could, however, statistically prove that Nkx 6.1-positive nuclei were absent from the latter mice. The extreme scarcity of these nuclei precluded the analyses of Nkx 6.1 mRNA levels by quantitative PCR. Only in adult animals could the latter method pick up detectable Nkx 6.1 mRNA levels, and this was paralleled by a profound increase in the number of Nkx 6.1-positive nuclei.
Our data therefore suggest that a normally functioning Pdx-1 gene is necessary for Nkx 6.1 expression. Interestingly, during pancreatic development we have found Pdx-1 and Nkx 6.1 immunoreactivities in the same developing parenchymal cells. During early development, staining for Nkx 6.1 was weaker than for Pdx-1, and this was parallelled by correspondingly lower mRNA levels (Øster et al. 1998). Therefore, it remains a tantalizing possibility that Pdx-1 may participate in the activation of the Nkx 6.1 gene which, subsequently, may act as a silencer of other genes.
Footnotes
Acknowledgements
Supported by the Danish National Research Foundation.